The federal government has recently adopted a prospective payment system for Medicare patients based on the Diagnosis-Related Group (DRG) patient classification system. Although considerable attention has been focused on the potential problems with the use of DRGs as patient categories, little attention has been paid to the potential impact of the methods used to calculate payment rates for each of the DRGs. This dissertation describes two separate artifacts in the Medicare rate-setting formulas that lead to discrepancies between total costs and total payments for each individual DRG and for all DRGs combined. The first artifact is due to the use of labor-market indexes to adjust hospital costs in different labor markets, while the second artifact is due to the use of hospital-weighted rather than patient-weighted average costs to calculate regional and national average payment rates. The discrepancies caused by these artifacts create an implicit redistribution of revenue -- i.e. a cross-subsidization -- between DRGs that has not been previously recognized or studied. The purpose of this dissertation is to analyze the economic incentives of case-mix hospital reimbursement, and to demonstrate how these incentives may be distorted by adjustment artifacts such as those in the Medicare prospective payment system. The specific goals of this dissertation are to: a) develop a decomposition-of-variance model to disaggregate DRG-based payment rates into separate incentive and artifact components, b) develop a goal-programming to remove the artifact components from Medicare payment rates, and c) compare the policy implications of adopting such an alternative schedule of payment rates with the implications of failing to remove the artifacts. The overall goal of this dissertation is to contribute to the understanding of how transfer payments can be calculated by government agencies so that consumers in different regions will have equal purchasing power without creating implicit patterns of cross-subsidization.